MobileNet V2

Use case : Image classification

Model description

MobileNetV2 improves upon V1 with inverted residual blocks and linear bottlenecks. It features skip connections between thin bottleneck layers, improving gradient flow and enabling deeper, more accurate networks.

The architecture uses inverted residuals that expand channels before depthwise convolution and then compress, with linear bottlenecks that remove non-linearity in narrow layers to preserve information. Skip connections between bottleneck layers improve gradient flow, typically with a 6x expansion ratio.

MobileNetV2 offers the best overall accuracy-efficiency trade-off with excellent quantization stability (typically <1% accuracy drop), making it ideal for production mobile applications, object detection backbones, and semantic segmentation networks.

(source: https://arxiv.org/abs/1801.04381)

The model is quantized to int8 using ONNX Runtime and exported for efficient deployment.

Network information

Network inputs / outputs

For an image resolution of NxM and P classes

Recommended platforms

Performances

Metrics

• Measures are done with default STEdgeAI Core configuration with enabled input / output allocated option.
• All the models are trained from scratch on Imagenet dataset

Reference NPU memory footprint on Imagenet dataset (see Accuracy for details on dataset)

Reference NPU inference time on Imagenet dataset (see Accuracy for details on dataset)

Accuracy with Imagenet dataset

Retraining and Integration in a simple example:

Please refer to the stm32ai-modelzoo-services GitHub here

References

[1] - Dataset: Imagenet (ILSVRC 2012) — https://www.image-net.org/

[2] - Model: MobileNetV2 — https://github.com/tensorflow/models/tree/master/research/slim/nets/mobilenet